RESUMEN
The ADI as a tool for risk management and regulation of food additives and pesticide residues is not readily applicable to inherent food plant toxicants: The margin between actual intake and potentially toxic levels is often small; application of the default uncertainty factors used to derive ADI values, particularly when extrapolating from animal data, would prohibit the utilisation of the food, which may have an overall beneficial health effect. Levels of inherent toxicants are difficult to control; their complete removal is not always wanted, due to their function for the plant or for human health. The health impact of the inherent toxicant is often modified by factors in the food, e.g. the bioavailability from the matrix and interaction with other inherent constituents. Risk-benefit analysis should be made for different consumption scenarios, without the use of uncertainty factors. Crucial in this approach is analysis of the toxicity of the whole foodstuff. The relationship between the whole foodstuff and the pure toxicant is expressed in the `product correction factor' (PCF). Investigations in humans are essential so that biomarkers of exposure and for effect can be used to analyse the difference between animals and humans and between the food and the pure toxicant. A grid of the variables characterising toxicity is proposed, showing their inter-relationships. A flow diagram for risk estimate is provided, using both toxicological and epidemiological studies.